Pulses of Deformation Reveal Frequently Recurring Shallow Magmatic Activity Beneath the Main Ethiopian Rift

Article Volume 12, Number 9 10 September 2011 Q0AB10, doi:10.1029/2011GC003662 ISSN: 1525‐2027

Pulses of deformation reveal frequently recurring shallow magmatic activity beneath the Main Ethiopian Rift

J. Biggs and I. D. Bastow Department of Earth Sciences, University of Bristol, Wills Memorial Building, Queen’s Street, Bristol BS8 1RJ, UK ([email protected]) D. Keir National Oceanography Centre, Southampton SO14 3ZH, UK E. Lewi IGSSA, Addis Ababa University, PO Box 1176, Addis Ababa, Ethiopia

[1] Magmatism strongly influences continental rift development, yet the mechanism, distribution, and timescales on which melt is emplaced and erupted through the shallow crust are not well characterized. The Main Ethiopian Rift (MER) has experienced significant volcanism, and the mantle beneath is characterized by high temperatures and partial melt. Despite its magma‐rich geological record, only one eruption has been historically recorded, and no dedicated monitoring networks exist. Consequently, the present‐day magmatic processes in the region remain poorly documented, and the associated hazards are neglected. We use satellite‐ based interferometric synthetic aperture radar observations to demonstrate that significant deformation has occurring at four volcanic edifices in the MER (Alutu, Corbetti, Bora, and Haledebi) from 1993 to 2010. This raises the number of volcanoes known to be deforming in East Africa beyond 12, comparable to many subduction arcs despite the smaller number of recorded eruptions. The largest displacements are at Alutu volcano, the site of a geothermal plant, which showed two pulses of rapid inflation (10–15 cm) in 2004 and 2008 separated by gradual subsidence. Our observations indicate a shallow (<10 km), frequently replenished zone of magma storage associated with volcanic edifices and add to the growing body of observations that indicate shallow magmatic processes operating on a decadal timescale are ubiquitous throughout the East African Rift. In the absence of detailed historical records of volcanic activity, satellite‐ based observations of monitoring parameters, such as deformation, could play an important role in assessing volcanic hazard.

Components: 6400 words, 6 figures, 2 tables. Keywords: InSAR; continental rifting; magmatic processes; volcano deformation. Index Terms: 1240 Geodesy and Gravity: Satellite geodesy: results (6929, 7215, 7230, 7240); 8105 Tectonophysics: Continental margins: divergent (1212, 8124); 8485 Volcanology: Remote sensing of volcanoes (4337). Received 15 April 2011; Revised 15 June 2011; Accepted 15 June 2011; Published 10 September 2011.

Biggs, J., I. D. Bastow, D. Keir, and E. Lewi (2011), Pulses of deformation reveal frequently recurring shallow magmatic activity beneath the Main Ethiopian Rift, Geochem. Geophys. Geosyst., 12, Q0AB10, doi:10.1029/2011GC003662.

Theme: Magma-Rich Extensional Regimes Guest Editors: R. Meyer, J. van Wijk, A. Breivik, and C. Tegner

1. Introduction intrusion expected to play a lesser role in accom- modating extension [Ebinger and Hayward, 1996].

[2] As a continental rift develops towards a new [5] Despite the numerous studies concerning the oceanic spreading center, extension initially accom- behavior and significance of magmatism on geo- modated in a broad zone of faulting and ductile logical timescales, little is known regarding the stretching transitions towards a narrower zone of short‐term activity of these volcanoes. Eruptions are focused magmatic intrusion [e.g., Ebinger and believed to have occurred in the 1800s at Fentale and Casey, 2001]. The spatial and temporal variations Kone [Rampey et al., 2010] and many others are in strain partitioning between faulting and magma- quotes as having undated lava flows with youthful‐ tism during progressive extension plays a vital role looking morphology [Siebert and Simkin, 2002]. in determining the continental rheology as well as The World Bank report on volcanic hazard lists local volcanic and seismic hazard. all these volcanoes at a zero level of monitoring and [3] Recent geophysical and geochemical experi- the highest level of uncertainty in terms of hazard ments in the Main Ethiopian Rift (MER) highlight and risk [Aspinall et al., 2011]. the importance of magmatism in achieving exten- [6] Here we present interferometric synthetic aper- sion during continental breakup [e.g., Keranen et al., ture radar (InSAR) observations from the seismi- 2004; Mackenzie et al., 2005; Bastow et al., 2011]. cally active MER (Figure 1), which encompasses The 60 km wide rift valley is bound by steep, large the transition between mechanical and magmatic offset, border faults that accommodated extension extension during the breakup of a continent [e.g., during the Miocene [Woldegabriel et al., 1990; Ebinger and Casey, 2001; Wolfenden et al., 2004]. Wolfenden et al., 2004]. Since the Quaternary, strain We investigate the spatial and temporal character- has localized to discrete, ∼20 km wide zones of istics of deformation and explore its source location, volcanism and small offset faulting where extension depth and geometry before placing the observations occurs predominantly through magma intrusion in the context of global volcanism, rift development, with little crustal thinning [e.g., Ebinger and Casey, geohazard and geothermal potential. 2001; Mackenzie et al., 2005; Maguire et al., 2006]. This narrow zone of localized strain in Quaternary magmatic segments is often termed the Wonji Fault 2. Geodetic Observations and Modeling Belt (WFB). Petrographic modeling, thermodynamic modeling and single clinopyroxene pressure esti- 2.1. Satellite Radar Archive mates have shown that the magmas predominantly [7] InSAR provides a powerful tool to measure fractionate in the upper crust [Rooney et al., 2011]. surface deformation associated with tectonic and Rhyolitic magmatism is associated with regularly magmatic processes across plate boundary zones. – spaced (20 50 km) central volcanoes of diameter The temporal and spatial characteristics of the >10 km, while basalts are usually associated with deformation patterns allow us to discriminate monogenetic vents or fissure eruptions along between dike intrusion, faulting and magma cham- magmatic segments [Mazzarini and Isola, 2010] ber pressure changes [e.g., Wright et al., 2006; (Figure 1). Pritchard and Simons, 2004; Massonnet et al., [4] Between 8 and 9.5°N, the along‐rift peak in 1993]. volume of Quaternary‐Recent volcanism [Abebe [8] Studies of surface deformation are frequently et al., 2007] coincides with a peak in seismic motivated by observations of seismic or eruptive anisotropy of the lithosphere and low seismic activity, in which case the location and timing of the velocities in the mantle [Bastow et al., 2010; Kendall event are already known. However, in this case, no et al., 2006]. Together, these observations suggest external information exists to guide our observations that the greatest melt supply, likelihood of volcanic and we perform a systematic search of all available eruption and geothermal potential lies between satellite radar data, extending back to 1993. The Gedemsa and Dofen volcanoes (Figure 1). South of Envisat background mission (2003–2010) acquired 8°N, geophysical constraints are sparse nonetheless three to four images per year over the MER. Before the distribution of seismicity [Keir et al., 2009], this, the archive is more limited: ERS1/2 collected Moho topography [Tiberi et al., 2005], combined images in 1993, 1997, 2000 and a few images are with geological observations [Ebinger et al., 2000] available form the Japanese Space Agency satellite suggest the deformation is more diffuse, with melt JERS between 1994 and 1996.

2of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

2.2. Sources of Deformation

[10] We identify four areas of significant deformation: (1) Alutu volcano; (2) Corbetti volcano; (3) the area between Bora Berrichio and Tulla Moje volcano (hereafter referred to as Bora), and (4) Haledebi volcano in the Angelele segment (Figure 4). At Alutu (1) and Corbetti (2) the circular deformation pattern is centered on the volcano (Figures 1c and 1d). Similar circular deformation at Bora (3) is located in a low‐lying area between the edifices of Bora Berrichio and Tulla Moje (Figure 1b). At Corbetti, four to five fringes of subsidence appear in two independent interferograms (ascending and descending tracks) over 1997–2000. Deformation extends beyond the patch of coherence associated with the edifice so the estimated displacement (14 cm) is a minimum. The JERS interferogram from 1994 to 1996 shows a small amount of uplift (1.4 cm) centered on the volcano but is dominated by a long‐wavelength ramp due to a lack of orbital information.

[11] The northernmost signal, Haledebi, is located Figure 1. Main Ethiopian Rift. Insets show example within the Angelele volcanic segment, >10 km south interferograms for (a) uplift at Haledebi (24 August of Hertali volcano. Although the deformation is 2007 to 11 November 2007), (b) uplift at Bora (27 August not associated with a catalogued volcano, a large 2008 to 23 June 2010), (c) uplift at Alutu (17 December volcanic edifice, with fresh‐looking lava flows, 2003 to 18 August 2004), and (d) subsidence at Corbetti is clearly visible in optical satellite images and (23 September 1997 to 13 September 2000). Each fringe digital elevation models. The deformation pattern represents 2.8 cm of motion in the satellite line‐of‐sight. is strongly asymmetric with a sharp boundary on the Volcanoes from the Smithsonian Database are denoted western side aligned with the local NNE‐SSW rift with black triangles or colored stars: green stars denote trend suggesting a structural control (Figure 1a). production plants, blue stars denote detailed exploration, and yellow stars denote prefeasibility studies. H, Hertali; D, Dofen; F, Fentale; B, Beru; K, Kone; B, Boset; 2.3. Source Models S, Sodore; G, Gedemsa; Bor, Bora Berricio; TM, Tulla Moje; EZ, East Ziway; A, Alutu; C, Corbetti. Squares [12] To estimate the dimensions and location of each mark the major cities of Addis Ababa, Nazret, and source (Figure 5 and Table 1) we use a nonlinear Awasa. Lakes are marked in pale blue. Dashed rectangles inversion to find the optimum parameters for a range are the coverage of Envisat Image Mode data from two of analytic solutions: a point source [Mogi, 1958], a tracks. uniform pressure, a penny‐shaped crack [Fialko et al., 2001] and a rectangular dislocation [Okada, 1985].

[9] Initially we produce annual interferograms [13] The best fitting source model for the deforma- (Figures 2 and 3) to identify areas of particular tion patterns at Alutu and Bora (Figure 6) is a shal- interest. We use the JPL software ROI_PAC and low (<2.5 km) penny‐shaped crack with large radius a 90 m SRTM DEM. For each, we process and (3–10 km), similar to the best‐fitting models for unwrap all available interferometric pairs and con- deformation at the Kenyan volcanoes [Biggs et al., struct a time series of deformation using a minimum 2009b]. While a deeper point source also produces norm velocity constraint [e.g., Berardino et al., a radially symmetric deformation pattern, the pre- 2002; Biggs et al., 2010]. Errors on each time step dicted line‐of‐sight displacements are asymmetric are estimated assuming a 1 cm error on each inter- due to the horizontal displacement component. ferogram (Figure 3); the cumulative error of <4 cm Thus, a shallower penny‐shaped model, which has is dominated by the early time steps [Biggs et al., larger ratio of vertical to horizontal displacement, 2010]. produces a more symmetric line‐of‐sight displace-

3of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

Figure 2. Survey interferograms from JERS, ERS, and Envisat for the central MER. Each color cycle (fringe) represents 2.8 cm of displacement in the satellite line‐of‐sight. The black dot is the summit, according to the Smithsonian Database [Siebert and Simkin, 2002]. Black rectangles highlight interferograms in which deformation is visible. ment pattern and always results in a lower root‐ deformation. The temporal characteristics of the mean‐square misfit. The deformation at Corbetti deformation, specifically that the direction of motion is better fit by a deeper (∼5 km) point source, but reverses, makes it unlikely that the deformation this may be due to the coherence‐limited coverage. can be explained by fault creep. Furthermore, the reversal of sign does not follow an annual pattern [14] The asymmetry in the deformation pattern at ruling out aquifer drainage and recharge. The pattern Haledebi indicates a significantly different geom- can be fit by a sill dipping at 10–20°W at a depth of etry to the other sites, and specifically, that struc- 2.7–8.8 km (Figure 5). tural features, such as faults, may influence the

4of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

Figure 3. Survey interferograms from JERS, ERS, and Envisat for the northern MER. Each color cycle (fringe) represents 2.8 cm of displacement in the satellite line‐of‐sight. The black dot is the summit, according to the Smithsonian Database [Siebert and Simkin, 2002]. Black rectangles highlight interferograms in which deformation is visible.

[15] For each example, the same source geometry provides acceptable fits to both periods of uplift and subsidence and the variation in parameters between interferograms is significantly smaller than Monte Carlo error estimates (Table 1) based on realistic noise distributions [e.g., Biggs et al., 2009b].

2.4. Temporal Characteristics

[16] The time series at Alutu shows two pulses – of rapid inflation (10 15 cm) in 2004 and 2008 Figure 4. Chart showing the results of the ground defor- separated by a period of more gradual subsidence mation survey at each volcano. Red denotes uplift; blue (Figure 6a). Periods of uplift began in 2008 and denotes subsidence; grey denotes no deformation; black 2009 at Bora and Corbetti, respectively (Figures 6b denotes no coherence; white denotes no interferogram and 6c). At Haledebi, the time series shows uplift available.

5of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

Figure 5. Example interferograms and models chosen to represent each period of significant displacement. Alutu inflation 1, 17 December 2003 to 18 August 2004; Alutu deflation 1, 22 September 2004 to 5 March 2008; Alutu inflation 2, 14 May 2008 to 1 December 2008; Alutu deflation 2, 14 January 2009 to 28 July 2010; Bora inflation, 26 December 2007 to 28 July 2010; Corbetti deflation (Desc), 23 September 1997 to 13 September 2000; Corbetti deflation (Asc), 28 September 1997 to 17 September 2000; Haledebi inflation, stacked rate for 8 interferograms between July 2007 and February 2008; Haledebi deflation, stacked rate for 122 interferograms between February 2008 and August 2010. Each fringe represents 2.8 cm of deformation. Model parameters are given in Table 1.

Table 1. Temporal and Spatial Characteristics of Observed Deformationa Displacement Length Decay Depth Radius Start (cm) (days) (days) Model (km) (km) Alutu Dec 2003 +15 260 230 Penny 0.7–2.5 2.8–8.9 Sep 2004 −4.71 400 320 Penny 0.5–1.9 2.9–10 Jul 2008 +9.9 150 230 Penny 0.7–2.4 4.0–8.2 Dec 2008 <−4.3 >750 320 Penny Bora Feb 2008 +5.3 900 1600 Penny 0.9–1.3 4.7–8.1 Corbetti 1994–1996b >1.4 ‐‐‐‐ ‐ 1997–2000b <14 ‐‐Mogi 5.8–7.8 0 ‐ −3.3 >2100 1500 ‐‐ ‐ Aug 2009 +4.2 290 1700 Mogi 3.3–5.3 0 Haledebi Jun 2007 +4.1 170 630 Sill 2.7–8.8 5.8 × 8 Dec 2007 −5.4 >1000 6800 Sill 2.7–8.8 5.8 × 8 aAll numbers are rounded to two significant figures. Errors in depth and radius reflect 1 sigma standard deviations from a log‐Gaussian distribution. bDates refer to the period of the interferogram rather than the source process.

6of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

2006]. Although the structure of hydrothermal and magmatic systems in the Main Ethiopian Rift are poorly characterized, the depths inferred from our InSAR models (Table 1) are consistent with the boundary between the hydrothermal and magmatic systems for a generalized, conceptual model. Deformation alone is not sufficient to discriminate between magmatic and hydrothermal processes so we must rely on geological constraints on the expected characteristics of each system.

[19] Prior to a 7.3 MW pilot geothermal plant Figure 6. Time series of deformation from Envisat opening in the Alutu‐Langano geothermal field in observations. (a) Alutu volcano showing two pulses of 1999, exploration drilling to 2.5 km encountered uplift in 2004 and 2008 with intervening phases of subsi- temperatures of 350°C. Extrusive products, includ- dence. (b) Haledebi showing uplift in 2007 followed by ‐ ing silicic tuffs, breccias, domes and obsidian flows, subsidence. (c) Bora Berricio showing two pulses of ∼ uplift in 2004 and 2008–2010 with intervening subsi- date from 155.8 ka to 2000 ya [e.g., Gianelli and dence. (d) Corbetti showing a pulse of uplift in 2009. Teklemariam, 1993; Gebregzabher,1986].The Color blocks are used to indicate periods of uplift rift‐parallel faulting places structural controls on (orange) and subsidence (green) that are clearly above the small‐scale, recent volcanic features (e.g., scoria noise levels. cones) and on the geothermal system [Gianelli and Teklemariam, 1993; Gizaw, 1993]. Fluids upwell of ∼4 cm from June to December 2007 followed by in a narrow, elongated zone along the fault, while subsidence (Figure 6d). inversion in the off‐axis temperature structure indicates lateral outflow at shallower depths (<1 km) [17] We paramaterize each period of deforma- [Gizaw, 1993]. Disequilibria between alteration tion using an exponential function described by start mineral assemblages, fluid inclusions and current date, magnitude and decay constant (Table 1). This well fluids [Gianelli and Teklemariam, 1993; Valori approach provides a framework for comparison et al., 1992] suggest the geothermal system is not between edifices. Pulses of inflation at Alutu last in steady state and is episodically perturbed. <1 year (∼9 and ∼5 months) with longer periods of subsidence between (3.8 and >2.1 years). The [20] The temporal pattern observed at Alutu, and decay constants are 7.5 months for inflation and to some extent, that at Bora and Corbetti, show 10.5 months for deflation. The magnitude of the remarkable similarities to that at Campi Flegrei, inflation (10–15 cm) is larger than that of defla- Italy where steady subsidence has been interrupted tion (4–5 cm) producing a steplike pattern with by pulses of uplift [Chiodini et al., 2003]. This overall inflation. The duration of uplift at Corbetti similarity is particularly surprising given the dif- (∼9 months) and Haledebi (∼6 months) is similar ference in tectonic setting and may suggest the to that at Alutu, but uplift at Bora occurred over volcanic processes responsible are independent of a significantly longer period (∼30 months). At the regional stress regime. At Campi Flegrei, uplift Corbetti and Bora, inflation and deflation are more may precede eruption (e.g., 7 m of uplift before 1538 linear (i.e., the decay constant is >4 years). Although Monte Nuovo eruption), but uplift alone does illustrated with an exponential fit, the time series not necessarily imply an eruption will follow (∼2m at Haledebi shows systematic variations in rate, of uplift was recorded in 1969–1972 and again which are greater than the estimated errors. in 1982–1984) [Troise et al., 2007]. The ratios of horizontal to vertical deformation are significantly different during uplift and subsidence: uplift is 3. Dynamics of Geothermal attributed to a deep (∼3 km) oblate source, and and Magmatic Systems subsidence to a shallower prolate source. Battaglia et al. [2006] present a model in which the intru- [18] Interpreting deformation patterns in terms of sion of magma takes place at the beginning of each magma movement is often ambiguous since in many period of unrest. Overpressure of magma or fluids cases deformation does not culminate in eruption of magmatic origin beneath an impermeable barrier [e.g., Biggs et al., 2009b], and conversely eruptions cause uplift. When a major breach of this zone also occur without deformation [e.g., Moran et al., occurs brine and gases migrated to the brittle, lower

7of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G pressure, colder aquifer. Faulting and brecciation Unfortunately the cracks lie in a vegetated region lead to efficient discharge of fluids by lateral migra- that is incoherent in the 3 year interferogram that tion resulting in subsidence. spans these events. Although Corbetti subsided by >14 cm of between 1997 and 2000, this is unlikely [21] A similar concept may be applicable to Alutu: to have directly caused cracks of the observed size magma‐driven uplift amplified by increased activity (up to 2.4 km by 2.5 m). However, Clark [1972] in the hydrothermal field, followed by subsidence proposed a fissure development mechanism by associated with the hydrothermal field alone. How- which small‐scale tectonic features are enlarged ever, a few crucial differences exist: (1) borehole during heavy rainfall provided it occurs before the evidence indicates that the fluid flow patterns of the initial cracks are refilled by aeolian sand, slumping geothermal reservoir are elongated parallel to the of the walls or sediment carried by light runoff major structural control whereas the deformation [Zellmer et al., 1985]. Applied to our observations, patterns are near circular, and (2) unlike Campi this implies that the underlying deformation was Flegrei, the periods of uplift and subsidence can continuous during 1996–1998, producing small‐ both be modeled by the same source. The shorter scale cracks above which fissures formed in the repeat times between pulses of uplift make Alutu an soil during heavy rainfall. Recent uplift at Corbetti ideal candidate for multidisciplinary studies of these following a 10 year hiatus in both deformation interactions, and future observations of this sys- and cracking, may echo the uplift observed between tem may be key to understanding the processes at 1994 and 1996 and could herald another period of coupled magmatic‐hydrothermal systems elsewhere. cracking and resulting disruption to infrastructure.

[24] Ethiopia, like Iceland, has high geothermal 4. Hazards and Resources potential: rifting above anomalously hot mantle in Continental Rifts promotes the emplacement of magma into the shallow crust. By detecting shallow magma storage [22] Continental rifting involves several processes driving hydrothermal circulation in locations such that represent a significant geohazard: dike intru- as Alutu we demonstrate the potential of InSAR sion, fault slip, silicic eruptions and ground cracking. as a prospecting tool for the geothermal industry. Furthermore, many events demonstrate interactions Previous exploration has mirrored the belief that between multiple processes [e.g., Wright et al., peak melt supply, geohazard and geothermal poten- 2006; Ayele et al., 2009; Biggs et al., 2009a; tial lie between Gedemsa and Dofen (Figure 1). Our Calais et al., 2008], thus magma migration beneath observations confirm that the Alutu‐Langano power the silicic centers may have implications for the plant is located above the most active zone of timing of dike intrusions, fault slip and ground magma migration in the MER and identify addi- cracking. In addition to the indicators of deep melt tional locations that warrant more detailed explora- supply between Gedemsa and Dofen already dis- tion (e.g., Corbetti, Bora). cussed, we see signs of shallow magma storage further south between Corbetti and Bora. Although we only have detailed information for the past few 5. Global Volcanism and Continental decades in the MER, the repeat times of either Deformation intrusive and extrusive events are expected to be at least an order of magnitude longer than the archive [25] Observations of shallow magma storage beneath of geodetic data (<20 years) presented here. Thus central volcanoes in the MER complement those we acknowledge that our study may not be charac- from Kenyan Rift [Biggs et al., 2009b], the Afar teristic of the long‐term behavior of the rift. None- Depression [Wright et al., 2006; Grandin et al., theless we demonstrate that the area of significant 2009; Keir et al., 2011], the Virunga volcanoes geohazard extends south beyond Gedemsa. [Wauthier et al., 2010] and Tanzania [Calais et al., 2008; Biggs et al., 2009a; Baer et al., 2008]. [23] Ground cracks (open fissures) are typically Together, these studies demonstrate that ∼yearlong observed following events such as earthquakes and inflation of shallow magma‐plumbing systems is dike intrusion, which involve significant deforma- a common characteristic of continental rifts. Of the tion [Asfaw, 1982; Rowland et al., 2007] and pose four observed deformation signals, two are directly a significant hazard to infrastructure. We investigate associated with well‐known volcanic edifices (Alutu possible links between cracks in Muleti village and Corbetti) and a third is located within an active during anomalously heavy rainfall in 1996–1998 area close to other volcanoes (Bora). However, unlike [Ayalew et al., 2004] and Corbetti <20 km away. observations from the Afar and Tanzania, where

8of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

Table 2. Comparison Between Volcano Records for MER (Alutu, Corbetti, Bora and Haledebi) between Different Tectonic Settingsa 1993 and 2010. This raises the number of volcanoes Total Historical Deformation Ratio known to be deforming in East Africa to 14, comparable to many arcs despite the smaller number E. Africa 110 20 14 8:1.4:1 Iceland 30 17 8 4:2.1:1 of recorded eruptions. Our observations indicate a Aleutians 91 42 15 6:2.8:1 shallow (<10 km), frequently replenished zone of Andes 175 62 18 10:3.4:1 magma storage associated with volcanic edifices Overall 406 141 55 7:2.5:1 and add to the growing body of observations that aThe total number of volcanoes and the number of these with indicate shallow magmatic processes operating on historical eruptions are taken from the Smithsonian catalog. The a decadal timescale are common throughout the number of volcanoes with deformation signals is taken from Fournier et al. [2010] and updated using this study. East African Rift. In the absence of detailed historical records of volcanic activity, satellite‐based observations of monitoring parameters, such as dike intrusion clearly accommodates a significant deformation, could play an important role in asses- component of the divergent plate motion [Wright sing volcanic hazard. et al., 2006; Grandin et al., 2009; Calais et al., 2008; Baer et al., 2008; Biggs et al., 2009a; Keir et al., 2011], the InSAR observations presented Acknowledgments here do not directly constrain the contribution of shallow magmatism to extension. [28] JB was funded by the European Space Agency Changing Earth Science Network, COMET+, and NERC New Investiga- [26] Compilations of magma fluxes show little variation between tectonic settings [White et al., 2006], tor grant NE/I001816/1; IB was funded by the Leverhulme Trust; DK was funded by NERC. We thank G. Demisee, yet there are differences in volcano deformation J. Gottsman, J. M. Kendall, and D. Pyle for useful discussions between arcs [Fournier et al., 2010]. We use the and the Editor Thorsten Becker, Cindy Ebinger, and one anon- updated record of volcano deformation, now 14 in ymous reviewer for their helpful comments. We thank ESA for East Africa, to compare a typical ocean‐ocean arc access to their archive of ERS and Envisat images. (Aleutians), a continental arc (Andes), an oceanic hot spot (Iceland) and a continental rift (E. Africa). Table 2 gives a comparison of three measures of References regional volcanic activity: (1) the total number of volcanoes [Siebert and Simkin, 2002]; (2) the Abebe, B., V. Acocella, T. Korme, and D. Ayalew (2007), Quaternary faulting and volcanism in the Main Ethiopian number of volcanoes with historical eruptions Rift, J. Afri. Earth Sci., 48(2–3), 115–124, doi:10.1016/j. [Siebert and Simkin, 2002], and (3) the number of jafrearsci.2006.10.005. volcanoes with known deformation signatures. The Asfaw, L. (1982), Development of earthquake‐induced proportion of deforming volcanoes varies from fissures in the Main Ethiopian Rift, Nature, 297, 393–395. 4:1 in Iceland to 10:1 in the Andes and the updated Aspinall, W., M. Auker, T. Hincks, S. Mahony, F. Nadim, J. Pooley, S. Sparks, and E. Syre (2011), Volcano hazard value for East Africa of 8:1 lies within this range. and exposure in GFDRR priority countries and risk mitiga- However, the ratio of volcanoes with historical tion measures, Volcano Risk Study 0100806‐00‐1‐R, Global eruptions ranges from 2.1 in Iceland to 3.4 in the Facil. for Disaster Reduct. and Recovery, Washington, D. C. Andes, but is only 1.4 in East Africa. Since the Ayalew, L., H. Yamagishi, and G. Reik (2004), Ground cracks proportion of historically active volcanoes is much in Ethiopian rift valley: Facts and uncertainties, Eng. Geol., 75, 309–324. lower in East Africa than in other locations, we Ayele, A., D. Keir, C. Ebinger, T. Wright, G. Stuart, W. Buck, attribute this difference to a reporting bias for his- E. Jacques, G. Ogubazghi, and J. Sholan (2009), September torical eruptions: the written record is shorter, and 2005 mega‐dike emplacement in the Manda‐Harraro nascent fewer deposits have eruption dates. Thus following oceanic rift (Afar depression), Geophys. Res. Lett., 36, this study, the number of volcanoes with recorded L20306, doi:10.1029/2009GL039605. Baer, G., Y. Hamiel, G. Shamir, and R. Nof (2008), Evolution deformation signatures for volcanoes in East Africa of a magma‐driven earthquake swarm and triggering of the is comparable to other regions and we see no sys- nearby Oldoinyo Lengai eruption, as resolved by InSAR, tematic variation between tectonic settings. ground observations and elastic modeling, East African Rift, 2007, Earth Planet. Sci. Lett., 272, 339–352. Bastow, I., S. Pilidou, J.‐M. Kendall, and G. Stuart (2010), 6. Conclusions Melt‐induced seismic anisotropy and magma assisted rifting in Ethiopia: Evidence from surface waves, Geochem. Geo- phys. Geosyst., 11, Q0AB05, doi:10.1029/2010GC003036. [27] We have demonstrated that pulses of deforma- Bastow, I., D. Keir, and E. Daly (2011), The Ethiopia Afar tion have occurred at four volcanic edifices in the Geoscientific Lithospheric Experiment (EAGLE): Probing

9of11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

the transition from continental rifting to incipient seafloor Gizaw, B. (1993), Aluto‐Langano geothermal field, Ethiopian spreading, in Volcanism and Evolution of the African Litho- rift valley: physical characteristics and the effects of gas sphere, edited by L. Beccaluva, G. Bianchini, and M. Wilson, on well performance, Geothermics, 22(2), 101–116, Spec. Pap. Geol. Soc. Am., 478,1–26, doi:10.1130/2011. doi:10.1016/0375-6505(93)90050-W. 2478(04). Grandin, R., et al. (2009), September 2005 Manda Hararo‐ Battaglia, M., C. Troise, F. Obrizzo, F. Pingue, and Dabbahu rifting event, Afar (Ethiopia): Constraints provided G. De Natale (2006), Evidence for fluid migration as the by geodetic data, J. Geophys. Res., 114, B08404, source of deformation at Campi Flegrei caldera (Italy), Geo- doi:10.1029/2008JB005843. phys. Res. Lett., 33, L01307, doi:10.1029/2005GL024904. Keir, D., I. Bastow, K. Whaler, C. Dg, and S. Hautot (2009), Berardino, P., G. Fornaro, R. Lanari, and E. Sansosti (2002), Lower crustal earthquakes near the ethiopian rift induced A new algorithm for surface deformation monitoring based by magmatic processes, Geochem. Geophys. Geosyst., 10, on small baseline differential SAR Interferograms, IEEE Q0AB02, doi:10.1029/2009GC002382. Trans. Geosci. Remote Sens., 40, 2375–2383. Keir, D., C. Pagli, I. D. Bastow, and A. Ayele (2011), The Biggs,J.,F.Amelung,N.Gourmelen,T.H.Dixon,and magma‐assisted removal of Arabia in Afar: Evidence from S. Kim (2009a), InSAR observations of 2007 Tanzania dike injection in the Ethiopian rift captured using InSAR rifting episode reveal mixed fault and dyke extension in an and seismicity, Tectonics, 30, TC2008, doi:10.1029/ immature continental rift, Geophys. J. Int., 179, 549–558, 2010TC002785. doi:10.1111/j.1365-246X.2009.04262.x. Kendall, J.‐M., S. Pilidou, D. Keir, I. Bastow, G. Stuart, and Biggs, J., E. Anthony, and C. Ebinger (2009b), Multiple infla- A. Ayele (2006), Mantle upwellings, melt migration and tion and deflation events at Kenyan volcanoes, East African the rifting of Africa: Insights from seismic anisotropy, Rift, Geology, 37, 979–982, doi:10.1130/G30133A.1. in The Afar Volcanic Province Within the East African Rift Biggs, J., Z. Lu, T. Fournier, and J. T. Freymueller (2010), System,editedbyG.Yirgu,C.J.Ebinger,andP.K.H. Magma flux at Okmok Volcano, Alaska, from a joint inver- Maguire, Geol. Soc. Spec. Pub., 259, 271–293. sion of continuous GPS, campaign GPS, and interferometric Keranen, K., S. Klemperer, and R. Gloaguen (2004), Three‐ synthetic aperture radar, J. Geophys. Res., 115, B12401, dimensional seismic imaging of a protoridge axis in the doi:10.1029/2010JB007577. Main Ethiopian rift, Geology, 32,949–952, doi:10.1130/ Calais, E., et al. (2008), Strain accommodation by slow slip G20737.1. and dyking in a youthful continental rift, East Africa, Nature, Mackenzie, G., H. Thybo, and P. Maguire (2005), Crustal 456, 783–787, doi:10.1038/nature07478. velocity structure across the Main Ethiopian Rift: results Chiodini,G.,M.Todesco,S.Caliro,C.DelGaudio, from two‐dimensional wide‐angle seismic modelling, G. Macedonio, and M. Russo (2003), Magma degassing as Geophys. J. Int., 162, 994–1006. a trigger of bradyseismic events: The case of Phlegrean Maguire, P., et al. (2006), Crustal structure of the northern Fields (Italy), Geophys. Res. Lett., 30(8), 1434, doi:10.1029/ main ethiopian rift from the eagle controlled‐source survey; 2002GL016790. a snapshot of incipient lithospheric break‐up, in The Afar Clark, M. (1972), Surface rupture along the Coyote Creek Volcanic Province Within the East African Rift System, fault, the Borrego Mountain earthquake of April 9, 1968, edited by G. Yirgu, C. J. Ebinger, and P. K. H. Maguire, U.S. Geol. Surv. Prof. Pap., 787,55–86. Geol. Soc. Spec. Pub., 259, 269–292. Ebinger, C. J., and M. Casey (2001), Continental breakup in Massonnet, D., M. Rossi, C. Carmona, F. Adragna, G. Peltzer, magmatic provinces: An Ethiopian example, Geology, 29, K. Fiegl, and T. Rabaute (1993), The displacement field of 527–530. the Landers Earthquake mapped by radar interferometry, Ebinger, C. J., and N. J. Hayward (1996), Soft plates and Nature, 364, 138–142. hot spots: Views from Afar, J. Geophys. Res., 101, Mazzarini, F., and I. Isola (2010), Monogenetic vent self‐ 21,859–21,876, doi:10.1029/96JB02118. similar clustering in extending continental crust: Examples Ebinger, C. T., D. Harding, S. Tesfaye, S. Kelley, and D. Rex from the East African Rift system, Geosphere, 6, 567–582. (2000), Rift deflection, migration, and propagation: Linkage Mogi, K. (1958), Relations between eruptions of various vol- of the Ethiopian and Eastern rifts, Africa, Geol. Soc. Am. canoes and the deformations of the ground surfaces around Bull., 112, 163–176. them, Bull. Earthquake Res. Inst., 36,99–134. Fialko, Y., M. Simons, and D. Agnew (2001), The complete Moran, S. C., O. Kwoun, T. Masterlark, and Z. Lu (2006), (3‐D) surface displacement field in the epicentral area of On the absence of InSAR‐detected volcano deformation the 1999 Mw7.1 Hector Mine earthquake, California, from spanning the 1995, 1996 and 1999 eruptions of Shishaldin space geodetic observations, Geophys. Res. Lett., 28(16), Volcano, Alaska, J. Volcanol. Geotherm. Res., 150,119–131, 3063–3066. doi:10.1016/j.jvolgeores.2005.07.013. Fournier, T. J., M. Pritchard, and S. Riddick (2010), Duration, Okada, Y. (1985), Surface deformation due to shear and magnitude, and frequency of subaerial volcano deformation tensile faults in a half‐space, Bull. Seismol. Soc. Am., 75, events: New results from Latin America using InSAR and 1135–1154. a global synthesis, Geochem. Geophys. Geosyst., 11, Pritchard, M. E., and M. Simons (2004), An InSAR‐based sur- Q01003, doi:10.1029/2009GC002558. vey of volcanic deformation in the central Andes, Geochem. Gebregzabher, Z. (1986), Hydrothermal alteration minerals in Geophys. Geosyst., 5, Q02002, doi:10.1029/2003GC000610. Aluto Langano geothermal wells, Ethiopia, Geothermics, Rampey, M. L., C. Oppenheimer, D. M. Pyle, and G. Yirgu 15(5–6), 735–740, doi:10.1016/0375-6505(86)90086-6. (2010), Caldera‐forming eruptions of the Quaternary Kone Gianelli, G., and M. Teklemariam (1993), Water‐rock inter- Volcanic Complex, Ethiopia, J. Afri. Earth Sci., 58,51–66, action processes in the Aluto‐Langano geothermal field doi:10.1016/j.jafrearsci.2010.01.008. (Ethiopia), J. Volcanol. Geotherm. Res., 56, 429–445, Rooney, T., I. Bastow, and D. Keir (2011), Insights into exten- doi:10.1016/0377-0273(93)90007-E. sional processes during magma assisted rifting: Evidence

10 of 11 Geochemistry Geophysics 3 BIGGS ET AL.: DEFORMING VOLCANOES IN THE MER 10.1029/2011GC003662 Geosystems G

from aligned scoria cones, J. Volcanol. Geotherm. Res., 201, Wauthier, C., V. Cayol, A. Hooper, F. Kervyn, P. Marinkovic, 83–96, doi:10.1016/j.jvolgeores.2010.07.019. N. D’Oreye, and M. P. Poland (2010), Activity of Rowland, J. V., E. Baker, C. J. Ebinger, D. Keir, T. Kidane, Nyiragongo and Nyamulagira Volcanoes (Dem. Rep. of J. Biggs, N. Hayward, and T. J. Wright (2007), Fault growth Congo) Revealed Using Geological, Geophysical and InSAR at a nascent slow‐spreading ridge: 2005 Dabbahu rifting data, Abstract G21D‐05 presented at 2010 Fall Meeting, episode, Afar, 171,1226–1246, doi:10.1111/j.1365- AGU, San Francisco, Calif., 13–17 Dec. 246X.2007.03584.x. White, S. M., J. A. Crisp, and F. J. Spera (2006), Long‐term Siebert, L., and T. Simkin (2002), Volcanoes of the World: An volumetric eruption rates and magma budgets, Geochem. Illustrated Catalog of Holocene Volcanoes and Their Erup- Geophys. Geosyst., 7, Q03010, doi:10.1029/2005GC001002. tions, Global Volcanism Program Digital Inf. Ser., GVP‐3, Woldegabriel, G., J. Aronson, and R. Walter (1990), Geology, Smithsonian Inst., Washington, D. C. [Available at http:// geochronology, and rift basin development in the central sec- www.volcano.si.edu/world/.] tor of the Main Ethiopia Rift, Geol. Soc. Am. Bull., 102, Tiberi, C., C. Ebinger, V. Ballu, G. Stuart, and B. Oluma 439–458. (2005), Inverse models of gravity data from the Red Sea‐ Wolfenden, E., C. Ebinger, G. Yirgu, A. Deino, and D. Ayalew Aden‐East African rifts triple junction zone, Geophys. J. (2004), Evolution of the northern Main Ethiopian Rift: birth Int., 163, 775–787, doi:10.1111/j.1365-246X.2005.02736.x. of a triple junction, Earth Planet. Sci. Lett., 224, 213–228. Troise, C., G. De Natale, F. Pingue, F. Obrizzo, P. De Martino, Wright, T., C. Ebinger, J. Biggs, A. Ayele, G. Yirgu, D. Keir, U. Tammaro, and E. Boschi (2007), Renewed ground uplift and A. Stork (2006), Magma‐maintained rift segmentation at at Campi Flegrei caldera (Italy): New insight on magmatic continental rupture in the 2005 Afar dyking episode, Nature, processes and forecast, Geophys. Res. Lett., 34, L03301, 442, 291–294, 2006. doi:10.1029/2006GL028545. Zellmer, J. T., G. Roquemore, and B. Blackerby (1985), Mod- Valori, A., M. Teklemariam, and G. Gianelli (1992), Evidence ern tectonic cracking near the Garlock Fault, California, of temperature increase of CO2‐bearing fluids from Aluto‐ Geol. Soc. Am. Bull., 96, 1037–1042. Langano geothermal field (Ethiopia): A fluid inclusions study of deep wells LA‐3 and LA‐6, Eur. J. Mineral, 4(5), 907–919.

11 of 11